From 4de8dd91044a18e51a2d87ebc5da979f7945c146 Mon Sep 17 00:00:00 2001 From: quenousimporte Date: Wed, 29 May 2024 21:35:13 +0200 Subject: [PATCH] le monde add author, compute date, attempt to add images --- lemonde.php | 59 ++++++++++++++++++++++++++++++++++++++--------------- 1 file changed, 42 insertions(+), 17 deletions(-) diff --git a/lemonde.php b/lemonde.php index fcbb75f..f8ef0ea 100644 --- a/lemonde.php +++ b/lemonde.php @@ -4,12 +4,13 @@ temp_folder = 'temp/'; - $epub->epub_file = $_POST['epub']; - $epub->title = $_POST['title']; + $epub->epub_file = 'lemonde' . $coverdate . '.epub'; + $epub->title = 'Le Monde ' . $coverdate ; - if ($_POST['cover']) - { - $coverurl = $_POST['cover']; - $coverdata = file_get_contents($coverurl); - file_put_contents('temp/cover.jpg', $coverdata); - $epub->AddImage( 'temp/cover.jpg', 'image/jpeg', true ); - } + $coverurl = 'https://www.lemonde.fr/thumbnail/journal/'. $coverdate .'/1000/1490'; + $epub->AddImage( $coverurl, 'image/jpeg', true ); // cache json in case url expires $tempjsonpath = 'temp/' . hash('md5', $url) . '.json'; @@ -50,18 +46,45 @@ foreach ($content as $article) { - $articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; })[0]; + $articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; }); + $articlebody = array_values($articlebody)[0]; + if ($articlebody->Title && $articlebody->HtmlText) { $pagecontent = '

'. strip_tags($articlebody->Title) . '

'; + + $author = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'author/xml'; }); + $author = array_values($author)[0]; + if ($author->Author) + { + $pagecontent .= $author->Author; + } + if ($articlebody->Introduction) { $pagecontent .= '' . $articlebody->Introduction . ''; } + + /*$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; })); + foreach ($images as $image) + { + $imageid = $image->ContentItemId; + $imageurl = preg_replace('/GetPublicationContentItems-.*\.json/', 'Image-MEDIUM-' . $imageid . '.jpg', $url); + + $tempcontent = file_get_contents($imageurl); + file_put_contents('temp/' . $imageid, $tempcontent); + + //$epub->AddImage('temp/' . $imageid, false, false ); + + $pagecontent .= '
'; + if ($image->HtmlText) + { + $pagecontent .= $image->HtmlText; + } + }*/ + $pagecontent .= $articlebody->HtmlText; $epub->AddPage($pagecontent, false, strip_tags($articlebody->Title)); - // echo('

'. strip_tags($articlebody->Title) . '

'); - // echo($articlebody->HtmlText); } } @@ -71,6 +94,10 @@ if ( ! $epub->error ) { echo 'Success: Download your book here.'; } + else + { + echo $epub->error; + } } else { echo $epub->error; @@ -80,11 +107,9 @@ ?>
-

Go to https://journal.lemonde.fr and hit f12 to get GetPublicationContentItems json and cover jpg urls.

+

Go to https://journal.lemonde.fr and hit f12 to get GetPublicationContentItems json url.

Json url:
- Cover url:
- Title:
- Epub file name:
+ Date: